JP-3734: Making OLS_C Default for CI Tests

[kmacdonald-stsci]

…ith CHARGELOSS.

Resolves JP-3734

Closes #

The JWST default algorithm was changed to be OLS_C for ramp fitting. The STCAL ramp fit testing has been changed to reflect this change in default algorithm.

Tasks

• update or add relevant tests
• update relevant docstrings and / or docs/ page
• Does this PR change any API used downstream? (if not, label with no-changelog-entry-needed)
  • write news fragment(s) in changes/: echo "changed something" > changes/<PR#>.<changetype>.rst (see below for change types)
  • run regression tests with this branch installed ("git+https://github.com/<fork>/stcal@<branch>")
    • jwst regression test
    • romancal regression test

news fragment change types...

• changes/<PR#>.apichange.rst: change to public API
• changes/<PR#>.bugfix.rst: fixes an issue
• changes/<PR#>.general.rst: infrastructure or miscellaneous change

[codecov]

Codecov Report

Attention: Patch coverage is 92.30769% with 2 lines in your changes missing coverage. Please review.

Project coverage is 73.79%. Comparing base (5f94030) to head (2be0844).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
src/stcal/ramp_fitting/ramp_fit.py	0.00%	2 Missing ⚠️

Additional details and impacted files

@@             Coverage Diff             @@
##             main     #289       +/-   ##
===========================================
- Coverage   85.19%   73.79%   -11.41%     
===========================================
  Files          46       46               
  Lines        8804     8809        +5     
===========================================
- Hits         7501     6501     -1000     
- Misses       1303     2308     +1005     

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[melanieclarke]

Looks fine to me! One very small clean up request.

Did you run regression tests somewhere?

[ddavis-stsci]

the romancal regression tests pass with this in place.
https://github.com/spacetelescope/RegressionTests/actions/runs/11034593835

[melanieclarke]

@kmacdonald-stsci - did regression tests get run for JWST for this change?

[kmacdonald-stsci]

@kmacdonald-stsci - did regression tests get run for JWST for this change?

Yes. None of the failures are related to any ramp fit tests. @tapastro verified the failures were due to another issue.

https://plwishmaster.stsci.edu:8081/job/RT/job/JWST-Developers-Pull-Requests/1742/#showFailuresLink

[tapastro]

It looks like this is breaking a unit test:

FAILED tests/test_ramp_fitting.py::test_neg_med_rates_single_integration_multi_segment_optional - AssertionError: 
  Not equal to tolerance rtol=1e-06, atol=0
  
  -inf location mismatch:
   ACTUAL: array([ 0.     , 60.53071,     -inf], dtype=float32)
   DESIRED: array([0., 0., 0.])

We should probably resolve this failure before merging.

[kmacdonald-stsci]

It looks like this is breaking a unit test:

FAILED tests/test_ramp_fitting.py::test_neg_med_rates_single_integration_multi_segment_optional - AssertionError: 
  Not equal to tolerance rtol=1e-06, atol=0
  
  -inf location mismatch:
   ACTUAL: array([ 0.     , 60.53071,     -inf], dtype=float32)
   DESIRED: array([0., 0., 0.])

We should probably resolve this failure before merging.

It's not clear to me this is caused by this PR. And I've never seen such a strange problem like this. If I run pytest tests/test_ramp_fitting.py the output to be checked for this test is different that if I run pytest tests/test_ramp_fitting.py::test_neg_med_rates_single_integration_multi_segment_optional.

It doesn't matter if I run with the OLS algorithm or the OLS_C algorithm. The screenshots of the test outputs show that the returned array is all zero when running the whole test script, but is non-zero if only the specific test is run, regardless of the choice of algorithm. Because this difference occurs independent of which algorithm is chosen, I don't think this PR can be the problem.

I'll continue to investigate.

[kmacdonald-stsci]

Closing this PR. This approach has introduced a subtle bug.